31/P magnetic resonance spectroscopy (MRS) of human cancers in vivo reveals metabolic characteristics which differ from those of normal tissues. These include elevations of phosphomonoesters (PME), phosphodiesters (PDE) and cellular pH, and diminished phophocreatine (PCr). Preliminary results in ongoing studies suggest that changes in PME's predict response of sarcomas to chemotherapy and that MRS-derived pH predicts response of soft tissue sarcomas to radiation and hyperthermia. The study we propose is a prospective trial in patients receiving neoadjuvant treatment of newly-diagnosed bone and soft tissue sarcomas. Attention will be paid to patient selection, protocol management, technical quality control and statistical analysis. State-of-the-art techniques will be used to optimize the metabolic information obtained. These include dual-tuned 1/H-/31P surface coils to access the wide variety of anatomic locations of sarcomas; 1/H decoupling of 31/P to distinguish individual components of phospholipid metabolites in the PME and PDE regions; image-guided 3-dimensional chemical shift imaging to accurately localize 31/P NMR spectra to tumor masses; and molar quantitation of metabolites. We will test the hypothesis that in vivo 31/P MRS can predict sensitivity or resistance to treatment by correlating response rate and disease-free survival to metabolic features in the baseline spectrum and to changes in the spectrum occurring early following initiation of treatment. Positive results in this study would pave the way for a clinical trial in which 31/P MRS would be used to intervene in the selection of treatment regimens in individual patients with sarcomas.